Virtual large arrays and lazy evaluation.
Design goals:
- Keep the public interface compact.
- Leverage standard Python syntax.
- Avoid overloading behaviour.
- Mimic NumPy when it doesn't contradict the other goals.
Use cases:
- Extract a lazy subset of a lazy array.
- Extract a sequence of concrete slices from a lazy array.
- MUST NOT make the full lazy array concrete.
- Stack a homogenous collection of lazy arrays to create a higher dimensional lazy array.
- Join a collection of compatible lazy arrays to create a larger lazy array of the same dimensionality.